Configure Nightwatch

You'll want this when your production trace volume gets expensive or you're capturing sensitive data you shouldn't be storing. The filtering flow is well explained: sampling controls which requests/commands start a trace, filtering drops specific event types after sampling, and redaction scrubs what you do keep. Includes practical patterns for route-based sampling (always sample checkout, never sample health checks), regex-based query filtering to ignore job table noise, and cache key patterns to skip session data. The three-stage pipeline diagram alone makes the sampling vs. filtering distinction clear, which most people get wrong initially.